The Amazonian savannas occupy approximately 150,000 km² of the Brazilian Amazon, occurring in scattered isolated patches over large areas of forest in the states of Amapá, Amazonas, Pará, Roraima and Rondônia. Despite having considerable variation in the Anuran composition between locations and between the savanna’s physiognomies, a systematic and geographically wide sampling has not been performed for the savanna from Amapá yet, located in the north of Brazil, eastern Amazonia. In this perspective, a study was conducted on the richness, composition, diversity, and abundance of Anuran species in a ​​savanna area in Amapá State. For Anuran sampling, we performed 24 samples in four physiognomies (grassland savanna, scrub grassland savanna, parkland savanna, open woodland savanna) through an active and auditory search more than 20 sampling plots of 100 × 50 meters in each physiognomy. Twenty-one (21) species of frogs belonging to five families were registered: Bufonidae, Hylidae, Leptodactylidae, Microhylidae and Phyllomedusidae. Scrub grassland savanna registered a greater number of individuals regarding the species richness by physiognomy. The species rarefaction curve for the total area reached an asymptote, suggesting that the data collection effort was enough to adequately sample the species richness of the area. The Kruskal-Wallis variance analysis revealed significant differences in the species richness and diversity among the physiognomies. The Bray-Curtis similarity analysis grouped the physiognomies into three main groups: open woodland savanna, grassland savanna and scrub grassland savanna and parkland savanna. Through ordering by non-metric multidimensional scaling, the species composition from the savanna anuran assemblage resulted in a separation among three sampled physiognomies with significant differences, indicating differences in assemblage composition of the three sampled physiognomies. The local richness (21 species) corresponds to 14% of the 15 typical species that have strongly associated distribution with the Cerrado from Central Brazil, and 35.6% of 59 typical species of neighboring domains which only marginally occur in the Cerrado, representing a considerable part of frog species richness recorded in the savanna in the eastern portion of the Brazilian Amazon.

Amapá, eastern Amazon, frog assemblages, tropical forest, amphibians

The Amazonian savannas are usually flat areas covered by open vegetation composed of herbaceous strata dominated by grasses, with trees and shrubs scattered in different coverage densities (Solbrig 1996). They occupy approximately 267,164 km² with almost 90% of the total area occurs in Bolivia and Brazil, occurring in isolated spots spread over large areas of forest in the states of Amapá, Amazonas, Pará, Roraima and Rondônia (Pires and Prance 1985; Sanaiotti et al. 1997) with smaller areas in Venezuela, Guyana and Suriname (Carvalho and Mustin 2017). The largest continuous savanna block are the Beni savannas in Bolivia (127,096 km²); the Guyanan savanna in Brazil, Venezuela and Guyana; the Sipaliwini-Parú savanna in Brazil and Suriname, and the savanna of Amapá in Brazil (13,027 km²) (Pennington et al. 2006, Barbosa et al. 2007, Carvalho and Mustin 2017).

The complexity and heterogeneity found in different savanna phytophysiognomies (Eiten 1972, Coutinho 1978), as well as the influence of neighboring domains, reveal the existence of geographic distribution patterns, diversity, richness and abundance of inhabiting species in these areas (Mesquita et al. 2006). With regard to fauna studies and given the great diversity of anuran amphibians from savannas, the number of studies about assemblage ecology in Amazonian savannas has been increasing over the years, addressing different aspects such as the mechanisms that act on the selection and use of habitats and the factors responsible for maintaining diversity, and in the proper assemblage structures (Menin et al. 2007; Menin et al. 2011; Rojas-Ahumada et al. 2012; Landeiro et al. 2014; Dias-Terceiro et al. 2015; Jorge et al. 2016; Ferrão et al. 2018; Ferreira et al. 2018).

In this context, despite the territorial extension and heterogeneity, the Amazon savannas are extremely unfamiliar areas in terms of their frog communities and are highly threatened by the expansion of human activity, therefore it is urgent to conduct inventories on anuran species (Neckel-Oliveira et al. 2000, Barreto et al. 2007, Bertoluci et al. 2007, Brasileiro et al. 2008, Giaretta et al. 2008, Pinheiro et al. 2012, Lima et al. 2017). The strong anthropic pressure, the high degree of endemism and declining savanna frog populations are other factors that justify the importance of inventories (Stuart et al. 2004, Eterovick et al. 2005).

Despite the considerable variation in the anuran composition between locations and between the Cerrado physiognomies (Neckel-Oliveira et al. 2000, Valdujo et al. 2012), a systematic and geographically wide study has not been performed for the Eastern Amazonia savanna. Therefore, an inventory on the anurans in the savanna area of Amapá was performed in order to know the composition, richness, and diversity along the different physiognomies. In this study, we investigated the following questions: (1) what is the species composition of anurans in the study area? (2) Do the species richness, diversity, and equitability differ between different types of savanna physiognomies? Additionally, we compared our findings with studies conducted in other Amazon savannas and well-sampled areas in the Cerrado in central Brazil.

The registered richness supports the estimated richness of 18-43 species in well sampled locations in the Cerrado of Central Brazil (Brasileiro et al. 2005, Vasconcelos and Rossa-Feres 2005, Santos et al. 2007, Brasileiro et al. 2008, Ribeiro-Jr and Bertoluci 2009, Araújo and Almeida-Santos 2011) and in Amazonian savannas (Neckel-Oliveira et al. 2000, Pinheiro et al. 2012, Bitar et al. 2012, Neckel-Oliveira et al. 2012, Lima et al. 2017). The local richness corresponds to 14% of the 150 species which have distribution strongly associated with the Cerrado, and 35.6% of the 59 typical species of neighboring domains and which only marginally occur in the Cerrado (Valdujo et al. 2011, 2012), representing 21 anuran species recorded in the savannas of the eastern portion of the Brazilian Amazon.

The anuran species richness of the studied savanna varied in different physiognomies, where the open formations (e.g., grassland savanna) had greater richness, followed by forest formations (e.g., parkland and open woodland savanna). This is probably a reflex of the proportion of abundant and rare species in each physiognomy (Magurran 2004). These data agree with other studies, which suggest that the open formations tend to have higher richness when compared to forest formations of the Domain Cerrado (Maffei et al. 2011). Studies carried out by Strüssmann et al. (2000), Brandão and Araújo (2002) and Valdujo et al. (2011) found higher anuran richness in open savanna formations than in the gallery forest. This is due to the fact that the reproduction of most savanna species occurs in open areas, and as the availability of breeding sites is one of the most important factors in the selection of habitat for anurans, higher species richness and abundance of individuals is expected in open physiognomies (Colli et al. 2002, Bastazini et al. 2007).

On the other hand, in the study performed by Neckel-Oliveira et al. (2000) in an area of the Amazonian savanna in the municipality of Santarém, by Pinheiro et al. (2012) in the Carajás region, both in the state of Pará, and by Lima et al. (2017) in the Amazonian Savanna of the Rio Curiaú Environmental Protection Area in Amapá, higher richness and abundance of anuran species were found in forested environments. Several studies report that species of open environments can use forested areas as sites for foraging, migration, reproduction, or refuge (Brassaloti et al. 2010, Oda et al. 2016, Santos and Conte 2016).

Furthermore, several studies have shown that complex and heterogeneous environments facilitate the coexistence of more species when compared to homogeneous environments (Conte and Rossa-Feres 2007, Santos et al. 2007, Vasconcelos et al. 2009, Silva et al. 2011). In addition to the heterogeneity and biodiverse of the Cerrado (Silva et al. 2006), differences in their composition and species richness between various sampled locations must have been favored by the contact with four major phytogeographic domains from South America: Amazonia, Atlantic Forest, Caatinga, and Chaco (Joly et al. 1999).

Despite the species cumulative curve having a tendency to stabilize, the possibility of local richness expansion is not unlikely, but as the study continues, an increased effort would contribute very slowly to adding to the species richness, as evidenced by richness estimators. This highlights the importance of accomplishing inventories with the association of different sampling methods for a more complete knowledge of anurans (Greenberg et al. 1994, Brown 1997, Maritz et al. 2007; Ribeiro-Jr et al. 2008), as the species richness is closely related to the sampling effort.

According to Santos (2006), accumulation curves are excellent for evaluating the inventories efficiency in the record of all species of certain sites or habitats. Similar to the rarefaction for the total area, the species accumulation curves for all physiognomies also stabilized; however, for the tree savanna stabilized in a relatively low species richness, this indicates that the additional sample effort would be necessary in the forest environments. In this context, three non-mutually exclusive factors and associated to the canopy cover structure may be responsible for the different richness patterns and frog species composition observed in the studied physiognomies: 1) some species ability in colonizing environments characterized by sparser vegetation; 2) the species distinct physiological tolerances concerning water temperature, light intensity and humidity near the soil surface; and 3) specific microenvironment dependence on specific sites for breeding.

In terms of diversity by physiognomies, the greatest diversity was found in scrub grassland savanna. The higher diversity recorded in the scrub grassland savanna corroborates comparative studies of the species diversity of open formations with grasslands backgrounds, savanna and forest in the Cerrado (Brasileiro et al. 2005, Araújo et al. 2009, Araújo and Almeida-Santos 2011, Valdujo et al. 2011). The diversity and equitability rates showed significant differences between sampled physiognomies. This result was expected due to the differences present in the physiognomies in the environmental heterogeneity (Vasconcelos et al. 2009, 2010), as this heterogeneity has an important role in determining species richness and in the assemblage structure by providing environments and several conditions for species with different ecophysiological requirements (Huston 1994, Buckley and Jetz 2007).

The obtained Bray-Curtis index seems coherent, presenting groupings that reflect differences between the main sampled physiognomies and revealing greater similarity between the grassland savanna and the scrub grassland savanna. Both physiognomies differ in the floristic composition of the dominant tree species, shrub and herbaceous substrate, and topography (Batalha 2011), and these differences have probably influenced the obtained results. However, geographically close areas generally exhibit similar structural features, which support similar assemblages in their richness and diversity (Dixo and Verdade 2006). Even though they do not favor species with more limited needs, the parkland savanna and open woodland savanna physiognomies, with lower structural complexity, are relatively favorable for structuring frog assemblages, thus potentially contributing to the preservation of regional diversity in mosaic vegetation areas (Gardner et al. 2006).

The Amapá savanna although scarcely known, may suffer from the agricultural expansion of grain production, extensive cattle ranching and urban growth, leading to habitat loss and vegetation fragmentation. Coupled with the wide diversity of anurans found in the area and the finding of new species and new records for Amapá State make Amapá savanna a hotpoint for anurans within the Amazon Forest hotspot (Silva and Bates 2002; Lima et al. 2017, Mustin et al. 2017) and, consequently, a place for the implementation of priority conservation measures aiming the increase of the protected area.

We are grateful to the Brazilian Agricultural Research Corporation – EMBRAPA/Amapá (Experimental Field) for allowing us access to research areas under their care and for logistical assistance. We thank the Fundação de Amparo à Pesquisa do Estado do Amapá/FAPEAP (process 250.203/059/2014). We thank Ananda Araújo and Telma Lobato for assistance in the field. We thank two anonymous reviewers for valuable suggestions that greatly improved this paper; and we are grateful the Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio) for providing collection permits (SISBIO/18243-1).

Voucher specimens

Voucher specimens are deposited at the Herpetological Collection of the Universidade Federal do Amapá (UNIFAP), under the care of Carlos Eduardo Costa Campos (CECCAMPOS).

BUFONIDAE

Rhinella major – CECCAMPOS 0747.

Rhinella sp. – CECCAMPOS 0799-0800, 0805-0806.

Rhinela marina – CECCAMPOS 0668.

HYLIDAE

Boana multifasciata – CECCAMPOS 0035, 0132, 0545.

Boana punctate – CECCAMPOS 0050, 0130, 0550, 0554, 0600.

Boana raniceps – CECCAMPOS 0486.

Dendropsophus cf. walfordi – CECCAMPOS 0765-0769.

Osteocephalus taurinus – CECCAMPOS 0676.

Scinax fuscomarginatus – CECCAMPOS 0037.

Scinax nebulosus – CECCAMPOS 0575-0580.

Scinax ruber – CECCAMPOS 0745, 0747, 0761.

Scinax x-signatus – CECCAMPOS 0321, 0683-0684.

Trachycephalus typhonius – CECCAMPOS 0739-0740.

LEPTODACTYLIDAE

Pseudopaludicola boliviana – CECCAMPOS 1153.

Adenomera hylaedactyla – CECCAMPOS 0047, 0548, 0560, 0609.

Leptodactylus fuscus – CECCAMPOS 0586, 0589, 0748, 0749, 0762.

Leptodactylus macrosternum – CECCAMPOS 0485, 0582.

Leptodactylus pentadactylus – CECCAMPOS 0670, 0737.

Leptodactylus podicipinus – CECCAMPOS 0048, 0049, 0606, 0753, 0772.

MICROHYLIDAE

Elachistocleis helianneae – CECCAMPOS 0122, 0227.

PHYLLOMEDUSIDAE

Pithecopus hypochondrialis – CECCAMPOS 0546, 0572, 0733.